257

ISO OBSERVATIONS OF FINE-STRUCTURE ATOMIC LINES FROM PROTO-PLANETARY NEBULAE

A. Castro-Carrizo1,D.Fong2, V. Bujarrabal1, M. Meixner2, A.G.G.M. Tielens3, W.B. Latter4,and M.J. Barlow5

1Observatorio Astron´omico Nacional, Apartado 1143, E-28800 Alcal´a de Henares, Spain 2University of Illinois, 1002 W. Green St., Urbana, Il.61801, USA 3Kapteyn Astronomical Institute, P.O. Box 800, 9700 AV Groningen, The Netherlands 4SIRTF Science Center/IPAC, CalTech, MS 314-6, Pasadena, CA 91125, USA 5Department of Physics and , University College London, Gower Street, London WC1E 6BT, United Kingdom

Abstract ejects a very fast and collimated wind that collides with We present ISO observations of fine-structure atomic the remnant of the AGB envelope. This phenomenom is lines from 24 evolved sources. Most of them are proto- thought to determine the morphology and dynamics of the planetary nebulae (PPNe) but we also include a few AGB PPN. and planetary nebulae. Data on O0,C+,N+,Si0, Both the increase of the temperature of the central Si+,S0,Fe0 and Fe+ were obtained. PPNe are found and the presence of shocks can originate the dissoci- to emit through low-excitation atomic lines only when ation of the molecules in the envelope. The photodissoci- the central star is hotter than ∼ 10000 K. This result ation regions (PDRs), that were theoretically studied by suggests that such lines predominantly arise from Photo- Tielens & Hollenbach (1985), are mainly cooled through Dissociation Regions (PDRs). Our results are also in rea- fine-structure atomic lines, the heating being due to the sonable agreement with predictions from PDR emission far-ultraviolet photons absorved by the dust grains. Those lines are excited collisionaly at temperatures of ∼ 102- models, allowing the estimation of the density of the emit- 3 ting layers from comparison with the model parameters. 10 K. On the other hand, the presence of shocks could However, Fabry-Perot ISO observations suggest in some play a role in the formation of relatively wide regions cases a contribution from shocked regions, in spite of their of atomic gas, which could also be cooled through fine- poor sensitivity and spectral resolution. The intensity of structure atomic lines (see Hollenbach & McKee 1989). the [C ii] 158 µm line has been used to measure the amount In this paper we analyze the results of our ISO data of low-excitation atomic mass in PPNe, since this tran- (that in some cases have been improved with data found sition has been found to be a useful model-independent in the ISO data archive or in Liu et al. 2001), and their probe to estimate the total mass of this component. In comparison with models in order to infer the origin of the most evolved sources the atomic mass is very high the emission of the low-excitation atomic gas. In Castro- (up to ∼ 1 M), representing in some cases the dominant Carrizo et al. (2001) and Fong et al. (2001) a detailed nebular component. description of the data, the comparison with models, and the calculation of the masses are found, for O-rich and Key words: Atomic data – Stars: AGB and post-AGB – C-rich sources respectively. (Stars:) circumstellar matter – Stars: mass-loss – (ISM:) planetary nebulae 2. ISO observations

Our sample is composed of 24 sources, mostly PPNe but also including a few planetary nebulae and AGB stars for 1. Introduction comparison. The observed O-rich sources are: OH 26.5+0.6, Mira, Betelgeuse, R Sct, AFGL 2343, HD 161796, 89 Her, In the last phases of the life of low mass stars (< 8 M), a M 1–92, M 2–9, Hb 12, Mz–3, and NGC 6302. The C-rich fast evolution from the Asymptotic Giant Branch (AGB) sources are: IRC +10216, IRAS 15194–5115, LP And, to planetary nebulae (PNe) takes place. AGB stars lose IRAS 22272+5435, AC Her, SAO 163075, AFGL 2688, Red most of their mass, forming a cool molecular envelope that Rectangle, IRAS 21282+5050, AFGL 618, NGC 6720, and expands isotropically at low velocity (∼ 15 km s−1). In less NGC 7027. We also observed off-source points to deter- than 1000 years these cool and extended AGB stars, and mine if interstellar medium (ISM) contribution exists, and their respective molecular envelopes, evolve becoming a in that case to estimate the emission that comes from the tiny and very hot blue dwarf surrounded by a mostly ion- . ized nebula. In the intermediate stage, the proto-planetary In order to study the low-excitation atomic gas we nebulae (PPNe) often show intermediate chemical and observed the following fine-structure atomic lines: [O i] physical properties between those of AGB envelopes and (63.2 µm, 145.5 µm), [C ii] (157.7 µm), [N ii] (121.9 µm), [Si i] PNe. At some point of this evolution the post-AGB star (68.5 µm, 129.7 µm), [Si ii] (34.8 µm), [S i] (25.2 µm), [Fe i]

Proc. Symposium ‘The Promise of the Herschel Space Observatory’ 12–15 December 2000, Toledo, Spain ESA SP-460, July 2001, eds. G.L. Pilbratt, J. Cernicharo, A.M. Heras, T. Prusti, & R. Harris 258 A. Castro-Carrizo et al.

Figure 1. Distribution in the H–R diagram of the observed sources.

(24.0 µm, 34.7 µm) and [Fe ii] (26.0 µm, 35.3 µm). We used detected. For a subsequent interpretation of the data it is both ISO spectrometers, LWS (43–196.7µm) and SWS also important to note that the number of detected lines (2.4–45 µm), according to different observational modes, and their intensities increase for sources surrounding hot- grating and Fabry-Perot (FP). ter central star. Only the FP observations allowed obtaining enough We have also found that the ratio of the most intense i ii spectral resolution to study the kinematics of the emitting lines, [O ]63µm/[C ] 158 µm, is very different when emis- sources, but due to the poor sensitivity of the FP modes sion comes from sources or from ISM contamination, being the analysis becomes very difficult. On the other hand, the < 1whenemissioncomesfromISMand> or 1whenit more sensitive grating modes do allow us to get reliable comes from the nebulae. This fact has helped us to infer line fluxes for the detected sources. (All our LWS-FP lines the origin of the detections in those few cases for which have been also observed by grating modes.) The typical we did not observed off-source points. rms obtained with the LWS/SWS grating is ∼ 0.1-1/1-10 −12 −2 −1 −1 10 erg cm s µm . The peak intensity of the most 3. Interpretation of the data intense detected lines is ∼10−8 erg cm−2s−1µm−1 for PNe, and ∼10−9-10−10 erg cm−2s−1µm−1 for PPNe. 3.1. Dependence on stellar parameters In Figure 1 we represent all the observed nebulae in One of the main results of this work is obtained by compar- an H-R diagram, showing the temperatures and the lumi- ison of the observed data with the stellar parameters, i.e., nosities of the central stars. The sources are marked with with the evolutionary stage of the sources. Figure 1 shows filled symbols when some of the observed atomic lines was that only nebulae surrounding stars hotter that ∼ 10000 K ISO Observations of Fine-Structure Atomic Lines from Proto-planetary Nebulae 259

are detected, and we have mentioned that as the Teff in- 3.3. PDR models creases above 10000 K the line emission is more intense and the number of lines detected is higher. The only excep- Our data have been compared with predictions of PDR tions are the detection of several lines in Betelgeuse and models. For the O-rich sources we used the last improved the probable detection [O i]63µm in Mira. Betelgeuse is a version of the models of Tielens & Hollenbach (1985), pub- red supergiant star which shows a well studied UV excess, lished by van den Ancker (1999). For the C-rich sources probably due to a very hot chromosphere, and Mira has new models have been developed by Latter & Tielens a hotter binary companion. Moreover, this dependence on (2001) taking into account the complex and different treat- the stellar temperature is strengthened by the fact that ment of the chemistry, what necessarily changes the phys- among emitters and non-emitters there are very similar ical conditions of the gas. The models solve the chemi- objects, from the point of view of their morphology, pres- cal and the thermal balance in the gas, and predict the ence of shocks, chemistry and total nebular mass, but with emission from these regions primarily as a function of the different low-excitation atomic emission and different Teff incident far-ultraviolet flux (G)andofthedensity(n)in of the central star (see for example the case of AFGL 2688 that region. Note the importance of the parameter G,that and AFGL 618). It is also noticeable the lack of emission represents the radiation field relevant for the photoelectric from the very massive molecular nebulae AFGL 2343 and effect, and which is the main PDR heater. HD 161796. The comparison of our data with models is quite satis- factory, since the physical parameters required to explain We interpret this result as showing that the total mass thedataarequitecompatibleforthedifferentlinesof of the low-excitation atomic gas strongly depends on the each source. In particular, the densities obtained from this temperature of the central star. This suggests that the analysis mostly range from 104 to 106 particles per cm−3. emitting region is a PDR caused by stellar UV photons, The main disagreement is that our PDR predictions do not due to shocks nor ISM photons. not account for the strong contrast found between the atomic emission of the nebulae around stars with more 3.2. Line profile analysis; Kinematics or less than ∼ 10000 K. The reason could be in the initial assumptions of the models. In the way we are using the theory, the only parameter of the models that depends Fabry-Perot profiles were obtained of M 2–9, IRAS 21282, on Teff is G. However, note that the number of photons AFGL 618, Hb 12, Mz–3, Hb 12, NGC 6720, NGC 7027 and deconvolution able to dissociate CO, which have between 6 and 11 eV, NGC 6302. After of the instrumental profile, dicreases considerably for the coolest stars, and that the we have analyzed the spectral profiles of those detected effective temperature assumed by the models is 30000 K. lines, in order to study the kinematics of the emitting This fact can lead to an overestimation of the dissociation regions. However, this analysis becomes quite uncertain capacity for the coolest sources. In addition, the model for our sources due to the mentioned poor sensitivity of assumption of chemical and physical equilibrium could be the FP observations. not satisfied for the coolest sources, and in that case this The main conclussion is that the widths of our decon- could contribute to overestimating the emission of those volved FP profiles are in general comparible to those of objects. < −1 low-velocity CO emission (∼ 40 km s ). Note that much higher velocities would be expected from shocked regions. 3.4. Shock models This suggests that atomic line emission mostly arises from PDRs. The shock theory used, for both J- and C-type shocks, predicts the intensity emitted through FIR lines from the However, in two particular cases, M 2–9 and AFGL 618, shock velocity and the pre-shock density. Those theoreti- part of the emission probably comes from shocks, since cal curves have been adapted from van den Ancker (1999) their profiles seem to show high-velocity expanding fea- for all the observed lines. The comparison of our data with tures (with a poor signal to noise ratio). For example, this theory of atomic line emission from shock excited re- whereas from the profile of the [O i]63µm line from a gions is less satisfactory than it was with the PDR theory. young PN, NGC 7027, we have found an expansion veloc- Different shock parameters, velocities and densities, are ity of ∼ 20 km s−1, to fit the profile of that emission com- needed to reproduce the intensities observed for the dif- ing from AFGL 618 we have needed to assume that there ferent lines of each source. In the case of J-type shocks, exists a gas component expanding at ∼ 70 km s−1.Higher the observed [C ii] intensities are too large for all models sensitivity and spectral resolution are needed to confirm taken into account, even those with very high shock ve- those shocked components, and in that case to determine locity (∼ 100 km s−1). For C-type shocks, models cannot the portion of mass accelerated. Also for our other sources explain the observed ionized atoms. Therefore, this analy- we expect to find minor atomic high-velocity components sis also suggests that the observed emission is not mainly with a more sensitive instrument. caused by shocks. 260

4. Atomic mass calculations source Matom Mmol Mion (M)(M)(M) Probably the best tracer of this low-excitation atomic re- < −4 −4 + Mira 210 1.5 10 O gions is [C ii] 158 µm, thanks to the high abundance of C Betelgeuse 2 10−4 410−4 O + in most of the PDR and to its easy analysis. Note that C IRAS 22272+5435 < 0.01 0.56 C,ISM appears almost at the same time that CO is photodisso- RSct < 0.002 0.002 O,ISM ciated, being the C0 region very thin for the O-rich case, AC Her < 0.01 1.1 10−4 C or coincident with the CO region (from the dissociation AFGL 2343 < 14.8O,ISM of other molecules) for the C-rich case. Moreover, C+ is HD 161796 < 0.05 0.68 O < soon photoionized in the H+ region. SAO 163075 0.001 0.03 C < The analysis of the [C ii] 158 µm line is easy because the AFGL 2688 0.02 0.6 C 89 Her < 0.005 0.0043 O following conditions are fulfilled. First, the total mass is −5 Red Rectangle < 0.002 2.5 10 C,ISM? independent of the excitation conditions, density and gas M 1–92 < 0.2 0.9 O,ISM temperature, because for the values expected the popula- M 2–9 0.04 0.005 0.004 O tion of the involved levels becomes constant. Secondly, we IRAS 21282+5050 0.1 2.7 0.008 C have checked that this emission in our cases is optically AFGL 618 0.01 1.6 4 10−4 C,ISM? thin. Hb 12 0.3 <0.001 0.015 O The mass of the low-excitation atomic gas is then given Mz–3 < 0.7 0.5 0.2 O,ISM by the equation 1, NGC 6720 0.18 0.34 0.2 C 6 −2 −1 2 NGC 7027 0.2 1.4 0.04 C M(M)=7.010 F[C II](erg cm s ) D(kpc) /χC (1) NGC 6302 1.3 0.1 0.2 O ii where F[C II] is the flux of the [C ] 158 µm line, D is Table 1. Low-excitation atomic gas masses derived from the distance to the source and χC is the abundance of [C ii] line flux. Mass estimates of the molecular region from C. For our calculations for O-rich sources, we have taken 12CO emission and of the ionized gas are given by comparison. −4 χC =310 , and for each C-rich source we have taken the Chemical information and the possible contribution of the ISM best value found in the bibliography. When [C ii] 158 µm are given in the last column. was not detected, we have estimated an upper limit. For the few cases in which we expect very low densities < 3 −3 (n ∼ 10 cm ) we must multiply this mass calculation by and Tielens acknowledge additional support from NASA grant a correction factor (> 1) that takes into account the de- 399-20-61 from the Long Term Space Astrophysics Program. pendence of the mass on the excitation conditions. Only for NGC 6720 and Hb 12, it has been necessary to include References this correction factor, with the values 2.5 and 1.5 respec- tively. Castro-Carrizo A., Bujarrabal V., Fong D., Meixner M., Tie- In Table 4 we show our calculations for the low-excitation lens A.G.G.M., Latter W.B. & Barlow M.J., A&A, in press. atomic gas (Matom) and, by comparison, the mass of the Fong D., Meixner M., Castro-Carrizo A., Bujarrabal V., Lat- molecular and the ionized gas found in the bibliography, ter W., Tielens A.G.G.M., Kelly D. & Sutton E., A&A, in press. considering the same distance values. In the last column Hollenbach D. & McKee C.F., 1989, ApJ 342, 306 we note the chemical type of each source and the presence Latter W.B. & Tielens A.G.G.M., in preparation. of ISM contamination. Liu X.-W., Barlow M.J., Cohen M., Danziger I.J., Luo S.-G. From the mass calculations we see that the atomic et al., 2001, MNRAS, in press. region increases when the sources evolve towards plane- Tielens A.G.G.M. & Hollenbach D., 1985, ApJ 291, 722 tary nebulae, because of photodissociation. However, it van den Ancker M.E., 1999, Ph.D. Thesis. is noticeable that the mass of the low-excitation atomic gas is relatively important for the most evolved sources, and even the most important component for Hb 12 and NGC 6302, becoming ∼ 1 M. Note also the particular cases of M 2–9 and 89 Her, for which a global deficiency of mass has been found. Another point to note is that the amont of atomic gas found does not seem to be related to the chemistry of the nebula. Acknowledgements Castro-Carrizo and Bujarrabal have been partially supported by the Spanish CYCIT, the European Commission and the PNIE under grants PB96-104, 1FD1997-1442 and ESP99-1291- E. Fong and Meixner have been supported by NASA JPL 961504, NASA NAG 5-3350 and NSF AST-97-33697. Latter